Today, when AI produces poor code, they no longer fix the code; they fix the system that produced it.

There is a powerful parallel in sales. For years, when sales execution wasn’t good enough, we tried to fix the salesperson. Train them. Coach them. Correct them. Send them on another programme. Listen to their calls and tell them what they should have said.

But what if the salesperson isn’t the right thing to fix – apart from the fact, that salespeople hate being “fixed”?

• If a salesperson struggles to explain your differentiation, improve the company’s answer.
• If several people struggle with the same buyer objection, improve the company’s guidance.
• If the wrong questions keep getting asked, improve the playbook.
• If people aren’t creating preference, improve the narrative they’re being equipped to use.

Fix the system that salespeople work in.

And AI makes this possible in a way that wasn’t practical before.

• Capture the know-how that lives in people’s heads and emerges in customer conversations.
• Turn it into company-approved guidance — better answers, questions, narratives and playbooks.
• Share what works — refine strong language and approaches discovered by one seller and distribute them across the team.
• Connect practice with execution — prepare before conversations, access guidance live, and practise afterwards.
• Improve the system at source — managers improve the guidance, so every improvement reaches the whole team.

Then the manager’s role changes.

• Instead of repeatedly correcting individual salespeople, the manager improves the company’s commercial system.
• Improve the answer once. Improve the question once. Improve the guidance once.
• And distribute that improvement across the team.

The salesperson still has to execute. They still have to learn, practise and exercise judgement. But they are no longer expected to invent the company’s expertise for themselves.